v0.8.13

🥳 Silero VAD Support

ML-based voice-activity detection vastly improves turn-end accuracy in noisy settings, enabling smooth conversations outdoors or at events.

• Add SileroVADProcessor for ONNX-based voice activity detection by @uezo in #431
• Add SileroVAD microphone button UI and logic by @uezo in #434
• Add preroll buffer to RecordingSession by @uezo in #430

🪄 TTS Pre-processing

Optional text pre-processing lets you fine-tune pronunciation (e.g., convert “OpenAI” to katakana) before synthesis.

• Add PreprocessText hook to SpeechSynthesizerBase by @uezo in #432

🤝 Grok & Gemini Compatibility

Removes OpenAI-specific params from the OpenAI-style endpoint, so Grok, Gemini, and other API-compatible models work out of the box.

• Add OpenAI compatible API option to ChatGPTService by @uezo in #433

🍩 Other changes

• Support ad-hoc internal request with idle state recovery by @uezo in #428
• Add support for language response from AIAvatarKit by @uezo in #429
• Fix bug where WebGL build fails by @uezo in #435

🎂 Birthday Release

This update drops on my birthday - thanks for celebrating with me! 🥳🎉🎈🍰

Full Changelog: v0.8.12...v0.8.13